A GOVERNMENT expert told an inquest yesterday that samples taken from a hydrogen generation plant hours before it exploded, killing two engineers, were non-representative because the various gases had separated. Mr Pang Sik-wing, a co-opted member of a government investigation team, said the hydrogen, being the lightest gas, was concentrated in the top of the plant while the other gases were at the bottom. Mr Pang, who is a senior officer in the Environmental Protection Department, said the team suspected that the sampling and analysing procedures employed by China Light and Power Company Ltd could have led to a misrepresentation of the actual composition. He was giving evidence at an inquest into the deaths of engineers Mr Wong Kwong-yu, 38, and Mr Yip Ka-pui, 40, who died in an explosion at Castle Peak Power Station on August 28 last year. He told the inquest he would have taken more samples to ensure there was no uneven mixing in the gases, adding that a more representative sample could have been taken from the bottom of the high-pressure storage receivers within the plant. Gas trapped in the pipeline leading to the top of the receivers should have been vented out before a gas sample was taken. He said if there had been insufficient venting, samples taken would not have been representative. He held that gas discarded by a chemist in this case was fairly marginal. The inquest was told that a chemist arrived at the power plant at 2.30 am and finished taking gas samples at 5 am, five hours and five minutes before the explosion. Results of the analysis, which indicated hydrogen purity in the receivers was about 98 per cent, were faxed to the power station at about 8.45 am. Mr Pang agreed it was a logical inference that hydrogen purity in the receivers would have been between 40 and 70 per cent at about 12.30 am. He said he had calculated the volume of gas inside pipes joining the sampling valve and receivers, based on information including vent diameter given by Mr Eric Cheung, the company's second chemist. But counsel for the company, Mr Nigel Kat, suggested the volume should be four times less than Mr Pang had calculated, because the actual vent diameter was half an inch rather than one inch. Coroner's officer Mr Patrick Li argued that, instead of claiming it was 100 per cent correct, Mr Pang's calculation had, in fact, reflected the worst possible scenario based on evidence given by Mr Cheung. It would be unfair to force Mr Pang to make calculations according to measurements which were not available to him at the time, he said. The hearing continues.